Description of problem: [ 762.208] Backtrace: [ 762.208] 0: /usr/bin/Xorg (xorg_backtrace+0x4a) [0x80abe5a] [ 762.209] 1: /usr/bin/Xorg (0x8047000+0x69dd8) [0x80b0dd8] [ 762.209] 2: (vdso) (__kernel_rt_sigreturn+0x0) [0xb773940c] [ 762.209] 3: /lib/libpixman-1.so.0 (pixman_image_get_width+0x6) [0x48bb7356] [ 762.209] 4: /usr/lib/xorg/modules/drivers/qxl_drv.so (0xb7612000+0x822b) [0xb761a22b] [ 762.209] 5: /usr/lib/xorg/modules/drivers/qxl_drv.so (0xb7612000+0x55a4) [0xb76175a4] [ 762.209] 6: /usr/bin/Xorg (0x8047000+0x10b90e) [0x815290e] [ 762.209] 7: /usr/bin/Xorg (0x8047000+0xe2c2e) [0x8129c2e] [ 762.209] 8: /usr/bin/Xorg (dixDestroyPixmap+0x14) [0x8072264] [ 762.209] 9: /usr/bin/Xorg (0x8047000+0x538cd) [0x809a8cd] [ 762.209] 10: /usr/bin/Xorg (FreeResource+0xec) [0x809b09c] [ 762.209] 11: /usr/bin/Xorg (0x8047000+0x2b51f) [0x807251f] [ 762.209] 12: /usr/bin/Xorg (0x8047000+0x3032d) [0x807732d] [ 762.209] 13: /usr/bin/Xorg (0x8047000+0x1dc35) [0x8064c35] [ 762.209] 14: /lib/libc.so.6 (__libc_start_main+0xf5) [0x48066635] [ 762.209] 15: /usr/bin/Xorg (0x8047000+0x1df79) [0x8064f79] [ 762.209] [ 762.209] Segmentation fault at address (nil) [ 762.209] Fatal server error: [ 762.209] Caught signal 11 (Segmentation fault). Server aborting I get this when switching users in my KVM machine. I reproduced in also in a freshly installed VM. It seems easily reproducible. Version-Release number of selected component (if applicable): xorg-x11-drv-qxl-0.0.22-0.fc17.i686 xorg-x11-server-Xorg-1.12.0-2.fc17.i686 How reproducible: always Steps to Reproduce: 1. log in 2. hit switch user, display switches to gdm 3. wait a while (few seconds, max minutes) 4. X restarts
Created attachment 579534 [details] Xorg log
Smolt: http://www.smolts.org/client/show/pub_38e3de8d-df77-4306-b80e-73cda92413d0
Created attachment 579536 [details] backtrace generated by abrt
Proposing as F17 Blocker.
When I switch from qxl driver to cirrus the VM does no longer crash.
While it is nasty, I don't think that this hits any of the F17 release criteria and I'm thinking this could be fixed with an update. Granted, it can't be fixed for the lives but I have a hard time seeing many people switch users in a livecd environment. I'm a weak -1 blocker on this
Discussed in the 2012-05-01 blocker bug review meeting. Rejected as a blocker for F17 final because it doesn't hit any of the F17 release criteria, is specific to VMs and could be fixed with an update.
Fix up - tflink left it blocking. -- Fedora Bugzappers volunteer triage team https://fedoraproject.org/wiki/BugZappers
Similar problem in bug #737924 - the reporter has done a bit of digging.
*** Bug 737924 has been marked as a duplicate of this bug. ***
Hans' comments from the F15 bug: Steps to Reproduce: 1. Take a 64 bit F-15 vm, using spice of course 2. Make sure it is fully up2date 3. Login to the fallback gnome desktop 4. switch to a text VT 5. Wait a bit Actual results: Xorg crashes, and restarts taking you back to the gdm login screen Expected results: Being able to work on the text VT and then return to the open gnome-session. Additional info: I've been debugging this and the problem is that src/qxl_surface.c: qxl_surface_kill() gets called on a surface with a NULL host_image, moreover &surface->cache->all_images[id] != surface, so this smells like memory corruption to me.
Cole: would https://bugzilla.redhat.com/show_bug.cgi?id=805017 be the same thing? Summary says 'Segmentation fault in anaconda when switching TTYs', but I've always just thought of it as 'the bug for when you switch to a VT in a VM and X crashes'.
Is http://article.gmane.org/gmane.comp.emulators.spice.devel/9534 intended to address this?
I don't think this was ever addressed in F17, and given that end-of-life is approaching I don't think it ever will be fixed. I haven't seen this issue on F18+, but if someone can reproduce on something newer, please reopen.